Delayed Rejection Adaptive Metropolis Algorithm Research Articles

Prediction of bridge structure deformation and strain based on dynamic testing and intelligent algorithms

The field load test is a direct and effective method for evaluating the performance of bridge structures. However, existing bridge field static load tests are costly and inefficient; moreover, they obstruct traffic and cause unavoidable damage to the bridge structure. As an alternative the the static load test, a random model update method based on bridge dynamic load tests and the Bayesian inference is proposed in this paper. The bridge static load test results were predicted with a high accuracy. To speed up the Bayesian method to infer the posterior probability density of the updated parameters, the Gaussian process was used in place of the finite element model, and the Bayesian inference used the Markov chain Monte Carlo method based on the delayed rejection adaptive Metropolis algorithm. First, the parameters to be modified for the bridge structure analysis model were determined based on the global sensitivity analysis method. Second, a uniform design sampling method was used to establish the Gaussian process optimization model to update the random model of the bridge structure. Finally, a reinforced concrete truss arch bridge was used to verify the correctness of the static load results of the bridge predicted by the random model update method based on dynamic load testing and Bayesian inference. The research results reveal that the prediction results of the bridge static load test based on the dynamic load test and Bayesian inference method agree with the actual test results.

Stochastic model updating of rotor support parameters using Bayesian approach

Uncertainties due to assembling, installation, and operational conditions are extensively involved in rotating systems' parameters. Stochastic characteristics of these parameters may seriously affect rotating systems' vibrational characteristics, such as their critical speeds and vibration amplitudes. These effects make the variability of parameters in the modeling of rotary machines inevitable. In rotating machinery, material properties and geometric parameters of the rotor, bearing characteristics and supports stiffness determine the system's dynamic response. Stochastic model updating methods consider model response variability and allocate them to the model parameters; however, they are not commonly employed in rotor dynamics, and deterministic approaches are still prevalent in this field. Due to the cost and efforts needed to set up experiments and obtain outcomes that reflect the machine's actual characteristics, stochastic updating practices of industrial rotating systems are rarely reported in the literature. This paper adopts an appropriate parameter selection procedure and suitable sampling strategy for stochastic model updating to investigate variability in the dynamic behavior of a complex turbo compressor rotor-bearing-support system, leading to successful parameter identification results. The compressor rotor is mounted on hydrodynamic journal bearings with speed-dependent stiffness and damping. Due to the rotating system complex model, a variance-based global sensitivity method is employed for parameter selection to eliminate non-influential parameters in the model updating and to alleviate updating complexity and computational burden. The Bayesian approach in the stochastic model updating is applied to estimate parameter uncertainty in the rotor with speed-dependent characteristics. Advanced Markov chain Monte Carlo sampling method using delayed rejection adaptive Metropolis algorithm is employed in the stochastic model updating. The updating procedure obtains marginal posterior probabilities of parameters, and uncertain parameter distributions are evaluated using the maximum entropy criterion.

Structural damage detection based on variational Bayesian inference and delayed rejection adaptive Metropolis algorithm

Existing studies on sparse Bayesian learning for structural damage detection usually assume that the posterior probability density functions follow standard distributions which facilitate to circumvent the intractable integration problem of the evidence by means of numerical sampling or analytical derivation. Moreover, the uncertainties of each mode are usually quantified as a common parameter to simplify the calculation. These assumptions may not be realistic in practice. This study proposes a sparse Bayesian method for structural damage detection suitable for standard and nonstandard probability distributions. The uncertainty corresponding to each mode is assumed as different. Variational Bayesian inference is developed and the posterior probability density functions of each unknown are individually derived. The parameters are found to follow the gamma distribution, whereas the distribution of the damage index cannot be directly obtained because of the nonlinear relationship in its posterior probability density function. The delayed rejection adaptive Metropolis algorithm is then adopted to generate numerical samples of the damage index. The coupled damage index and parameters in the variational Bayesian inference are successively calculated via an iterative process. A laboratory-tested frame is utilised to verify the effectiveness of the proposed method. The results indicate that the sparse damage can be accurately detected. The proposed method has the advantage of high accuracy and broad applicability.

Optimization Design of Groundwater Pollution Monitoring Scheme and Inverse Identification of Pollution Source Parameters Using Bayes’ Theorem

In the process of identifying groundwater pollution sources, in order to solve the problem that the monitoring data of monitoring wells was insufficient or the correlation between monitoring data and model parameters was weak, a monitoring well optimization method based on Bayesian formula and information entropy was proposed. Two-dimensional phreatic groundwater solute transport model was built and solved by using GMS software. To reduce the computational load of calling the numerical model repeatedly in the optimization design of the monitoring schemes and the identification process of the pollution sources, the Kriging method was used to establish the surrogate model of the numerical model. Under the condition of single well monitoring and determined monitoring frequency, with the target of optimization of monitoring position number D and monitoring time interval ∆t, both the single-objective monitoring scheme with the minimum information entropy of the model parameter posterior distribution and the multi-objective monitoring scheme with the minimum information entropy and the shortest monitoring time were optimized respectively. According to the above-optimized monitoring schemes, the delayed rejection adaptive Metropolis algorithm was used to identify the pollution source parameters. The case study results showed that under the condition of pre-set single well monitoring with monitoring frequency of 10 times, the single-objective optimized monitoring scheme was D = 37 and Δt = 20 days. Under this monitoring scheme, the mean errors of inversion pollution source parameters α = (XS, YS, T1, T2, QS) were 0.09%, 0.4%, 4.72%, 2.43%, and 9.29%, respectively. The multi-objective optimized monitoring scheme was D = 37 and Δt = 2 days. Under this monitoring scheme, the mean errors of the inversion parameters α = (XS, YS, T1, T2, QS) were 12.76%, 3.77%, 5.13%, 1.36%, and 7.68%, respectively. Compared with the monitoring scheme based on the single-objective optimization, although the inversion mean error of the five parameters based on the multi-objective optimized monitoring scheme increased by 2.75%, the monitoring time significantly reduced from 180 to 18 days.

Reliability of Poisson-Nernst-Planck Anomalous Models for Impedance Spectroscopy.

We investigate possible connections between two different implementations of the Poisson-Nernst-Planck (PNP) anomalous models used to analyze the electrical response of electrolytic cells. One of them is built in the framework of the fractional calculus and considers integro-differential boundary conditions also formulated by using fractional derivatives; the other one is an extension of the standard PNP model presented by Barsoukov and Macdonald, which can also be related to equivalent circuits containing constant phase elements (CPEs). Both extensions may be related to an anomalous diffusion with subdiffusive characteristics through the electrical conductivity and are able to describe the experimental data presented here. Furthermore, we apply the Bayesian inversion method to extract the parameter of interest in the analytical formulas of impedance. To resolve the corresponding inverse problem, we use the delayed-rejection adaptive-Metropolis algorithm (DRAM) in the context of Markov-chain Monte Carlo (MCMC) algorithms to find the posterior distributions of the parameter and the corresponding confidence intervals.

Accelerating MCMC via Kriging-based adaptive independent proposals and delayed rejection

Markov Chain Monte Carlo (MCMC) simulation has a considerable computational burden when the target probability density function (PDF) evaluation involves a black-box, potentially computationally-expensive, numerical model. A novel framework to accelerate MCMC is developed here for such applications. It leverages a Kriging surrogate model approximation to the target PDF to improve computational efficiency, while preserves convergence properties to the exact target PDF, avoiding potential accuracy problems introduced through the surrogate model error. The approach relies on the delayed-rejection (DR) scheme and combines the rapid exploration characteristics of global (independent) proposals with the local search robustness of random walk proposals under the MCMC setting. The global proposal is chosen as the Kriging-based target PDF approximation. This proposal may resemble, depending on the surrogate model error, the actual target PDF very well, and therefore can provide significant computational benefits, including fast convergence of the Markov chain to its stationary distribution and, more importantly, low correlation between samples. At each MCMC step a candidate sample is generated from the independent proposal. If rejected, DR allows an extra random walk around the current sample. The DR step guarantees convergence to the actual target PDF, circumventing robustness problems that may arise when the Kriging-based independent proposal offers a poor approximation to (i.e., underestimates) the actual target PDF. The overall computational efficiency is further improved through an adaptive updating of the Kriging surrogate model during the MCMC sampling phase by extracting information from candidate samples whose inclusion in the training database can substantially enhance Kriging’s accuracy. The computational efficiency and robustness of the established algorithm, termed Adaptive Kriging Delayed Rejection Adaptive Metropolis algorithm(AK-DRAM), are demonstrated in two analytical benchmark problems and two engineering problems focusing on conditional failure sample simulation and Bayesian inference.

Prediction of soil water retention curve using Bayesian updating from limited measurement data

• A new updating algorithm based on the Bayesian framework predicts soil water retention curve with limited measured data. • Prediction algorithm is effective to reduce the uncertainty of parameters and soil water retention curve prediction. • Reliable soil water retention curve can be obtained using prediction algorithm even with incomplete prior knowledge. A soil water retention curve is one of the fundamental elements used to describe unsaturated soil. The accurate determination of soil water retention curve requires sufficient available information. However, the amount of measurement data is generally limited due to the restriction of time or test apparatus. As a result, it is a challenge to determine the soil water retention curve from limited measurement data. To address this problem, a Bayesian framework is proposed. In the Bayesian framework, Bayesian updating can be employed using the posterior distribution that is obtained by the Markov chain Monte Carlo sampling method with the Delayed Rejection Adaptive Metropolis algorithm. The parameters of soil water retention curve model are represented by the sample statistics of updating posterior distribution. A new updating algorithm based on Bayesian framework is proposed to predict the soil water retention curve using the ideal data and the limited measurement data of the granite residual soil and sand. The results show that the proposed prediction algorithm exhibits an excellent capability for more accurately determining the soil water retention curve with limited measured data. The uncertainty of updating parameters and the influence of the prior knowledge can be reduced. The converged results can be derived using the proposed prediction algorithm even if the prior knowledge is incomplete.

Model selection for dynamic reduction-based structural health monitoring following the Bayesian evidence approach

Abstract There usually exist multitudinous finite element (FE) models with varying level of complexity which can be developed from the engineering judgment for the purpose of structural system identification and health monitoring. By following the theory of Bayesian evidence statistic, this paper proposes a methodology to investigate the issues of FE model-class selection for choosing suitable parameterized structural models utilized in dynamic reduction-based structural health monitoring (SHM). By employing the concept of information divergence, the amount of information needed to be extracted from the measured data is explicitly quantified during the procedure of FE model updating-based structural health monitoring. Then, for achieving a trade-off between the complexity of a parameterized FE model class and that of its corresponding information-theoretic interpretation, such information is utilized for penalizing the complexity of model class to ensure that a relatively simple parameterization scheme can be achieved. The proposed methodology consists of calibration and subsequent monitoring stages, and the information obtained in the former stage is utilized as pseudo-data which is learned by the latter stage to improve the model parameter estimation by implementing the delayed rejection adaptive Metropolis algorithm. Through numerical case studies conducted for a four-storey two-bay steel frame structure considering semi-rigid connections as well as laboratory experiment performed for a two-storey bolt-connected steel frame model, the feasibility and validity of proposed methodology is demonstrated.

Quantified Validation with Uncertainty Analysis for Turbulent Single-Phase Friction Models

This paper introduces a framework for model selection that includes parameter estimation, uncertainty propagation, and quantified validation. The framework is applied to single-phase turbulent friction modeling in CTF, which is a thermal-hydraulic code for nuclear engineering applications. The friction model is chosen because it is well understood and easy to separate from other physics, which allows focus to be on the model selection framework instead of on the particulars of the chosen model. Two different empirical models are compared: the McAdams Correlation and the Simplified McAdams Correlation. The parameter estimation is performed by calibrating each of the friction models to experimental data using the Delayed Rejection Adaptive Metropolis algorithm, which is a Markov Chain Monte Carlo method. State point uncertainties are also considered, which are determined based on measurement errors from the experiment. The input parameter distributions are propagated through CTF using a statistical method with samples. A variety of validation metrics is used to quantify which empirical model is more accurate. It is shown that model form uncertainty can be quantified using validation once all other sources of uncertainty—numerical, sampling, experimental, and parameter—have been quantitatively addressed. When multiple models are available, the one that has the smallest model form error can be selected. Though the framework is applied to a simple example here, the same process can quantify the model form uncertainty of more complicated physics, multiple models, and simulation tools in other fields. Therefore, this work is a demonstration of best practices for future assessments of model form uncertainty.

Calibration of ADM1 using the Monte Carlo Markov Chain for modeling of anaerobic biodigestion of sugarcane vinasse in an AnSBBR

The Anaerobic Digestion Model No. 1 (ADM1) was applied in this study to simulate the processing of sugarcane vinasse in an Anaerobic Sequencing Batch Biofilm Reactor (AnSBBR). In the model calibration, the values of the free parameters, concentration of anaerobic biomass for specific species (Xsu, Xc4, Xpro, Xac, Xh2)11Xsu: Sugar degraders, Xc4: Butyric and valeric acids degraders, Xpro: Propionic acid degraders, Xac: Acetic acid degraders, Xh2: Hydrogen degraders, were obtained by minimizing the objective function expressed by the determinant of the Box-Draper Matrix. Afterwards, these parameters were optimized with the Monte Carlo Markov Chain Method, using the Delayed Rejection Adaptive Metropolis Algorithm. The Geweke statistic test evaluated the convergence of the method. All results passed for direct and cross validation to elucidate the methane production, production and consumption of the organic acids and organic matter removal. The ADM1 was able to show the syntrophism between the species and to describe the experimental data, thus making it possible to be applied in others scenarios. However, it was not so adequate for all conditions in cross validation due to a lack of experimental data and the need to include other metabolic pathways in the model. The chosen concentration of anaerobic biomass for specific species were shown as an important step in this work, because these parameters are difficult to be determined in practice.

MCMC methods for inference in a mathematical model of pulmonary circulation

This study performs parameter inference in a partial differential equations system of pulmonary circulation. We use a fluid dynamics network model that takes selected parameter values and mimics the behaviour of the pulmonary haemodynamics under normal physiological and pathological conditions. This is of medical interest as it enables tracking the progression of pulmonary hypertension. We show how we make the fluids model tractable by reducing the parameter dimension from a 55D to a 5D problem. The Delayed Rejection Adaptive Metropolis algorithm, coupled with constraint non‐linear optimization, is successfully used to learn the parameter values and quantify the uncertainty in the parameter estimates. To accommodate for different magnitudes of the parameter values, we introduce an improved parameter scaling technique in the Delayed Rejection Adaptive Metropolis algorithm. Formal convergence diagnostics are employed to check for convergence of the Markov chains. Additionally, we perform model selection using different information criteria, including Watanabe Akaike Information Criteria.

On Estimating Non-standard Discrete Distributions Using Adaptive MCMC Methods

In this paper, we compare empirically the performance of some adaptive MCMC methods, that is, Adaptive Metropolis (AM) algorithm, Single Component Adaptive Metropolis (SCAM) algorithm and Delayed Rejection Adaptive Metropolis (DRAM) algorithm. The context is the simulation of non-standard discrete distributions. The performance criterion used is the precision of the frequency estimator. An application to a Bayesian hypothesis testing problem shows the superiority of the DRAM algorithm over the other considered sampling schemes.

Statistical Inference of The Mechanisms Driving Collective Cell Movement

SummaryNumerous biological processes, many impacting on human health, rely on collective cell movement. We develop nine candidate models, based on advection–diffusion partial differential equations, to describe various alternative mechanisms that may drive cell movement. The parameters of these models were inferred from one-dimensional projections of laboratory observations of Dictyostelium discoideum cells by sampling from the posterior distribution using the delayed rejection adaptive Metropolis algorithm. The best model was selected by using the widely applicable information criterion. We conclude that cell movement in our study system was driven both by a self-generated gradient in an attractant that the cells could deplete locally, and by chemical interactions between the cells.

Laplace Variational Approximation for Semiparametric Regression in the Presence of Heteroscedastic Errors

Variational approximations provide fast, deterministic alternatives to Markov chain Monte Carlo for Bayesian inference on the parameters of complex, hierarchical models. Variational approximations are often limited in practicality in the absence of conjugate posterior distributions. Recent work has focused on the application of variational methods to models with only partial conjugacy, such as in semiparametric regression with heteroscedastic errors. Here, both the mean and log variance functions are modeled as smooth functions of covariates. For this problem, we derive a mean field variational approximation with an embedded Laplace approximation to account for the nonconjugate structure. Empirical results with simulated and real data show that our approximate method has significant computational advantages over traditional Markov chain Monte Carlo; in this case, a delayed rejection adaptive Metropolis algorithm. The variational approximation is much faster and eliminates the need for tuning parameter selection, achieves good fits for both the mean and log variance functions, and reasonably reflects the posterior uncertainty. We apply the methods to log-intensity data from a small angle X-ray scattering experiment, in which properly accounting for the smooth heteroscedasticity leads to significant improvements in posterior inference for key physical characteristics of an organic molecule.

Thermoacoustic modeling and uncertainty analysis of two-dimensional conductive membranes

A model for two-dimensional graphene-based thermoacoustic membranes is investigated analytically and numerically validated using Bayesian statistics in this study. The temperature and the pressure variables are first analytically determined in one-dimension by noticing that the magnitude of the pressure time derivative is small in the heat transfer equations and by taking advantage of the large disparity between the length scales. The one-dimensional findings are then extended to three-dimensions, where pressure fluctuation produced by the surface temperature variation is determined using an acoustic piston model. Through the one and three-dimensional model analysis, the dependence of acoustic pressure as a function of frequency is studied. The acoustic response with respect to the frequency shows different characteristics when assuming Dirichlet (temperature) or Neumann (heat flux) boundary conditions. The thermoacoustic model is validated with a graphene-on-paper loudspeaker using Bayesian statistical methods and a Delayed Rejection Adaptive Metropolis algorithm to identify model parameters and their uncertainty. The findings provide insights into the heat transport mechanisms associated with sound generation from thermally cycling thin conductive membranes at high frequencies.

Quantification of parameter and model uncertainty for shape memory alloy bending actuators

In this article, we employ a Bayesian framework to estimate parameter and model uncertainty for shape memory alloy bending actuators. The Bayesian framework provides parameter densities, instead of ordinary least-squares optimal point estimates. Bayes’ rule relates a posterior parameter density to a prior density and likelihood. However, the posterior density is difficult to calculate directly for high-dimensional parameter spaces. Markov chain Monte Carlo methods overcome this difficulty indirectly by creating a Markov chain whose stationary density is the posterior. In this article, we utilize the Delayed Rejection Adaptive Metropolis algorithm for estimating parameter uncertainty. The shape memory alloy bending actuator is modeled using the homogenized energy framework, a computationally efficient and accurate model for various transductive materials. The model is summarized, and techniques for estimating the heat transfer parameters are presented. An algorithmic approach to quantifying uncertainty is useful for numerous reasons. The anticipated use is to quantify uncertainty for robust control algorithms. Robust control is an area of considerable research for smart materials such as shape memory alloy; however, the source of uncertainty is rarely quantified. The methods employed here would greatly aid in the design of robust controllers.

Advanced Markov Chain Monte Carlo Approach for Finite Element Calibration under Uncertainty

AbstractUncertainty involved in the experiment data prohibits the wide applications of the finite element (FE) model updating technique into engineering practices. In this article, the Markov Chain Monte Carlo approach with a Delayed Rejection Adaptive Metropolis algorithm is investigated to perform the Bayesian framework for FE updating under uncertainty. A major advantage of this algorithm is that it adopts global and local adaptive strategies, which makes the FE model updating be robust to uncertainty. Another merit of the studied method is that it not only quantitatively predicts structural responses, but also calculates their statistical parameters such as the confidence interval. Impact test data of a grid structure are investigated to demonstrate the effectiveness of the presented FE model updating technique, in which the uncertainty parameters include the vertical and longitudinal spring stiffness that simulate the boundary conditions, the end‐fixity factor for modeling semi‐rigid connections, and the elastic modulus for simulating the uncertainty associated with material property.

The Polya Tree Sampler: Toward Efficient and Automatic Independent Metropolis–Hastings Proposals

We present a simple, efficient, and computationally cheap sampling method for exploring an unnormalized multivariate density on ℝd, such as a posterior density, called the Polya tree sampler. The algorithm constructs an independent proposal based on an approximation of the target density. The approximation is built from a set of (initial) support points—data that act as parameters for the approximation—and the predictive density of a finite multivariate Polya tree. In an initial “warming-up” phase, the support points are iteratively relocated to regions of higher support under the target distribution to minimize the distance between the target distribution and the Polya tree predictive distribution. In the “sampling” phase, samples from the final approximating mixture of finite Polya trees are used as candidates which are accepted with a standard Metropolis–Hastings acceptance probability. Several illustrations are presented, including comparisons of the proposed approach to Metropolis-within-Gibbs and delayed rejection adaptive Metropolis algorithm. This article has supplementary material online.